DE4226749C2 - Method for determining variables that characterize driving behavior - Google Patents

Description

The invention relates to a method for determining the driving behavior characteristic quantities according to the generic term of Claim 1.

A linear single-track model of a vehicle is known from which neglects the height of the center of gravity of the vehicle becomes. Thus, in this approximation, the focus of the driving stuff in the plane of the contact points of the wheels. There thus roll and pitch movements are excluded, can this model the wheels of an axle to a wheel in the middle of the axis. This model is for example in DE-Buch: Zomotor, Adam: Chassis technology, driving behavior, ed. Jörnsen Reimpell, Würzburg: Vogel 1987, ISBN 3-8023-0774-7 pages 99 to 116.

This representation does not show how driving behavioral characteristic quantities depending on measured whose sizes can be determined.

From DE 36 08 420 C2 it is known to measure the vehicle longitudinal speed, Vehicle lateral acceleration and yaw rate to measure and calculate a float angle to use. The properties are used for the calculation of the vehicle itself taking into account the vehicle model used.

The object of the invention is to provide a method for determining the To design driving characteristics of characteristic quantities in such a way that the best possible measurement accuracy with a possible ge effort required hardware can be achieved.  

This task is in a generic method for loading parameters that characterize the driving behavior according to the invention with the characteristic features of the An solved claim 1, wherein the features of the subclaims Mark partial training and further education.

The method according to the invention shows advantages in that that input variables in the form of control and fault inputs are not must be known that no vehicle parameters are needed that and that both small and large float angles can be estimated.

The present invention relates to a method for determining the driving behavior characterizing quantities, in which sensors are installed in the vehicle, the longitudinal acceleration a _x and the lateral acceleration a _{y of} the vehicle in the center of gravity, the yaw rate Ω _z and the vehicle speed in the longitudinal direction v _x directly measure up. The vehicle speed in the transverse direction v _y and / or the slip angle β can then be determined from these variables. The relationship applies:

β = arctan (v _y / v _x ) (1).

In the following, a model is shown in which the vehicle speed in the transverse direction v _{y is} determined. The float angle β can then be determined from this in accordance with equation (1). This model is based on the fact that the speed components are coupled via the rotational speeds around the longitudinal, vertical and transverse axes.

According to DIN 70 000, the following characterize the movement differential equations:

dv _x / dt + Ω _{y *} v _z - Ω _{z *} v _y = F _x / m (2)

dv _y / dt + Ω _{z *} v _x - Ω _{x *} v _z = F _y / m (3)

dv _z / dt + Ω _{x *} v _y - Ω _{y *} v _x = F _z / m (4)

and

I _{xx *} dΩ _x / dt + (I _zz - I _yy ) _* Ω _{y *} Ω _z + M _x (5)

I _{yy *} dΩ _y / dt + (I _xx - I _zz ) _* Ω _{x *} Ω _z = M _y (6)

I _{zz *} dΩ _z / dt + (I _yy - I _xx ) _* Ω _{y *} Ω _x = M _z (7)

The quantities F _x , F _y and F _{z are} the forces which act in the direction corresponding to the index, the quantities M _x , M _y and M _z the moments about the axes indicated by the index, the quantities I _xx , I _yy and I _{zz are} the moments of inertia with respect to the axles indicated by the indices and the size m is the vehicle mass.

To simplify further mathematical treatment now proposed a linear state model. Doing so exposed that the rotational speeds Ω measured accurately can be. In the matrix representation, this results in a Sy stem of differential equations:

The last summand of equation (8) (F _x / m, F _y / m, F _z / m) ^T can be expressed by the measured acceleration signal (a _x , a _y , A _z ) ^T and the gravitational acceleration g:

This results in a linear differential equation with respect to the velocity components (v _x , v _y , v _z ) ^T :

The angles Γ (pitch), Φ (roll) and Π (yaw) are gimbals Position angle and describe the transformation of the geodesic Coordinate system in the vehicle-fixed coordinate system.

Further simplifications of the above model result if one assumes that the vehicle is on one plane (that is, that the cardan-bearing angles are negligible), if the components v _x and v _{y are} considered to be longitudinal and transverse speeds (ie , if the influence of the construction angle is neglected). If there are no pitching or rolling movements of the vehicle, the terms Ω _y * y _z , Ω _y * v _x , Ω _x * v _z , Ω _x * v _{y are} all equal to 0. The system of equations (10) is then simplified :

In principle, it is possible to obtain the state variables (v _x , v _y ) ^T from equation (11) by integration. However, the instability of equation (11) can lead to errors.

Because the speed in the longitudinal direction v _{x can be} measured, the speed in the transverse direction v _y can be observed. The observer design for v _{y is} given below.

The differential equation (11) has the following general ones State space representation:

dx / dt + A (t) x + u (t) (12)

The following applies to the associated measurement equation:

y = c ^T _* x (t) = (10) ^T x (t) + v _x (13)

This general approach of a system of equations corresponds to the present model:

Through the known transformation into the normal observation form is obtained from equations (12) and (13) full observer of the form:

d x / dt + (A (t) - k (t) _* c ^T ) t)) x + k (t) _* y + u (t) (15)

with the measurement equation

y + c ^T _* x (t) = (10) ^T x (t) + v _x (16)

The underlined vectors relate to the representation in the normal observation form. The quantity k (t) is the observer gain k (t) = (k ₁ , k ₂ ) ^T. If you explicitly write out equation (15), you get:

d x ₁ / dt = x ₂ _* Ω _z + k₁ _* (y- x ₁) + a _x (17)

d x ₂ / dt + - x ₁ _* Ω _z + k₂ _* (y- x ₁) + a _y (18)

The determination of the observer gain k (t) is state of the art Technology as a method by Prof. O. Föllinger: "Design time-variant systems due to pole specification ". This method is shown published in the International Journal of Control, 1983 Volume 38 No. 2, Pages 419 to 431 in the article by Bestle and Zeitz: Canonical form observer design for non-linear time-variable systems; there in particular on p. 421. It follows from this literature put the so-called Luebberger observer of the form:

The sizes p ₀ and p _{1 are} freely selectable as coefficients of the characteristic polynomial.

Alternatively, it is possible to use the observer gain of a Kalman filter. One such method is in the book by Brammer / Siffling: "Kalman-Bucy-Filter" in the series "Methods of Control Engineering" by R. Oldenbourg in Munich, Vienna from 1985. It follows the following system of equations consisting of the equations (20), (21), (22), (23):

dp₁₁ / dt = -2 _* p₂₁ _* dΩ _z / dt - p₂₁ _* / R₁₁ + Q _k11 (20)

dp₂₁ / dt = -p₂₂ _* dΩ _z / dt + p₁₁ _* dΩ _z / dt - p₂₁ _* p₂₂ / R₁₁ + Q _k21 (21)

dp₂₂ / dt = 2 _* p₂₂ _* dΩ _z / dt - p₂₂ _* / R₁₁ + Q _k22 (22)

and

The values of the coefficients of the covariance matrix are Q _k11 = 1, Q _k21 = 0.3, Q _k22 = 1 and the value R ₁₁ = 0.5. The initial values p ₁₁ (0) = 0, p ₂₁ (0) = 0 and p ₂₂ (0) = 0.

In an expansion of the model, the effect of the gimbal angles can be compensated for, which were neglected in the previous consideration. For this purpose, a subsystem 1 is first described in which Ω _z and v _z are 0. The following equation then results:

dv _x / dt = a _x + sin (Γ _* g (24)

The time mean of the quantity sin (Γ) _* g can thus be calculated from the difference between dv _x / dt and a _x if Ω _z and v _z are 0.

The variables v _x and v _y can then be determined by means of a second subsystem using the variable sin (Γ) * g determined at Ω _z = 0. This subsystem then has the equations:

dv _x / dt = Ω _{z *} v _y + a _x + sin (Γ) _* g (25)

dv _y / dt = -Ω _{z *} v _x + a _y + µ (26)

dµ / dt = 0 (27)

Then Ω _x and Ω _{y are} not required. The size µ is taken into account separately and corresponds to the expression -cos (*) * sin (*) * g. Either an observer or a Kalman filter is then used for this subsystem.

Alternatively, it is also possible to use all expressions that are little contain at least one of the angles Γ or Φ as an error and expand the order of the system. It then turns out for example the following system of equations, which again can be treated with a Kalman filter:

dv _x / dt = v _{y *} Ω _z + a _x + e₁ (28)

dv _y / dt = -v _x * Ω _z + a _y + e₂ (29)

de₁ / dt = 0 (30)

de₂ / dt = 0 (31)

and

y = v _x (32)

An embodiment of the invention is in the drawing is shown schematically and is described in more detail below. Show it:

Fig. 1 shows the block diagram of an observer, in which the pitch and roll angles Γ and Φ were neglected in the underlying model and

Fig. 2 shows an arrangement of sensors with which the measured variables on which the models are based are determined.

As can be seen from the block diagram in FIG. 1, the vehicle speed in the longitudinal direction v _x , the longitudinal acceleration a _x , the transverse acceleration a _y and the yaw rate Ω _{z are used as} measured variables. The circles represent summation points, with the rectangles with the point the input quantities are multiplied by each other. The integrators and amplifiers are self-explanatory.

The block diagram of FIG. 1 is an illustration of the anti-tions (17) and (18).

Fig. 2 shows certain signals that are supplied by sensors known per se to a computing device 407 . The measurement quantities that correspond to the signals 401 , 402 , 403 , 404 , 405 and 406 are shown in FIG. 4. In the computing device, for example, the vehicle speed in the transverse direction v _{y is} determined on the basis of the method described at the outset and output as signal 408 . From this signal, the float angle β can then be calculated in the computing unit 409, for example by means of the equation (1). This value is then output as signal 410 .

Claims (4)

1. A method for determining the variables that characterize the driving behavior, with a computing device ( 407 ) to which signals ( 401 , 402 , 403 , 406 ) are supplied, the measured variables of the longitudinal vehicle speed (v _x ), the longitudinal acceleration (a _x ), and the lateral acceleration (a _y ) and the yaw angular velocity (Ω _z ), and in which at least one further variable is derived on the basis of these measured variables, characterized in that the further variable is derived using the measured variables and state equations, and that as derived Size at least the float angle (β) is determined and output. 2. The method according to claim 1, characterized in that the equations of state are transformed into the normal observation form and that the at least one further variable ( 408 , 410 ) is derived by means of a complete observer. 3. The method according to claim 1, characterized in that the observer gain ( 17 , 18 ) is derived by means of a Kalman filter. 4. The method according to claim 1, 2 or 3, characterized in that in the state equations roll and pitch movements of the vehicle are compensated in their effect ( 404 , 405 ).